JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 97, NO. D18, PAGES 20,421-20,426, DECEMBER 20, 1992 Comparison of General Circulation Models to Earth Radiation Budget Experiment Data: Computation of Clear-Sky Fluxes ROBERTD. CESS, 1 GERALD L. POTTER, 2 W. LAWRENCE GATES, 2 JEAN-JACQUES MORCRETTE, 3 AND LISA CORSETTI 2 The recent availability of top-of-the-atmosphere radiometric measurements from the Earth Radia- tion Budget Experiment providesimportant opportunities for testing and improving numerical climate models. What is unique about these satellite data is that they provide monthly mean clear-sky measurements.There is, however, considerable confusion as to evaluating clear-sky radiative fluxes in climate models in a manner that is consistent with the satellite data processing system. This study provides a clear-sky flux computation method that serves as an analog to the data processing procedure and so provides a model diagnosticthat is consistentwith the processedsatellite data. 1. INTRODUCTION The availability of data from the Earth Radiation Budget Experiment (ERBE) has provided unique insights concern- ing the radiative impact of clouds on the Earth-atmosphere system. This was accomplished by separately averaging clear-sky, top-of-the-atmosphere (TOA) shortwave and longwave radiative flux measurements, so that the TOA radiation budget with clouds present could be compared to that without clouds. It was found, for example, that clouds radiatively cool the Earth-atmosphere system on a global average, because their enhancement of shortwave reflection to space (i.e., cooling) is stronger than their longwave (i.e., greenhouse) warming [Rarnanathan et al., 1989]. Hemi- spherically, however, the shortwave and longwave contri- butions nearly cancel each other in the winter hemisphere, and it is for the summer hemispherethat shortwave cooling dominates [Harrison et al., 1990]. Because the net TOA radiation acts to heat the summer hemisphere and cool the winter hemisphere, then clouds tend to reduce the seasonal change in net TOA radiative heating. In addition to improving our physical understanding of cloud-radiative impacts, the ERBE data provide an impor- tant means of testing such interactions within atmospheric general circulation models (GCMs). However, there is con- siderableconfusion as to evaluating clear-sky TOA fluxes in GCMs in a manner that is consistent with the elaborate ERBE data processing system [Potter et al., 1992]. The purpose of the present study is to demonstrate a procedure for clear-sky flux computation that is representative of the ERBE processing while at the same time being relatively straightforward to implement in a GCM. To demonstrate this, we employ the European Centre for Medium-Range Weather Forecasts (ECMWF) GCM. lInstitute for Terrestrial and Planetary Atmospheres, State Uni- versity of New York, Stony Brook. 2Lawrence Livermore National Laboratory, Livermore, Califor- nia. 3European Centre for Medium-Range Weather Forecasts, Read- ing, Berkshire, England. Copyright 1992 by the American Geophysical Union. Paper number 92JD01726. 0148-0227/92/92JD-01726505.00 2. ERBE DATA PROCESSING It will be useful to briefly review the ERBE data process- ing procedure as pertains to clear-sky fluxes. The most comprehensive ERBE data sets are those that combine scanner measurements from two satellites: the Earth Radi- ation budget Satellite (ERBS) and NOAA 9. The ERBS is in a 57ø inclination, 600-km altitude orbit which, counting both ascending and descending nodes, provides sampling of all local hours every 36 days. NOAA 9 is in an 870-km altitude sun-synchronous orbit with nominal equator crossing times at 0230 and 1430 local time. The scanner pixels are roughly circular in shape with nadir diameters of about 35 km for ERBS and 50 km for NOAA 9. The scanner measurements consist of radiances that are converted to fluxes through use of angular-directional models [Brooks et al., 1986]. These instantaneous fluxes are then diurnally averaged by employ- ing diurnal models [Brooks et al., 1986], and this diurnal averaging is performed for both cloudy (all sky) and clear conditions. It is thus important to recognize that the ERBE clear-sky fluxes comprise diurnal averages, even though clear conditions might exist for only a small fraction of the day. These daily fluxes are then used to produce monthly means for 2.5 ø x 2.5ø grids. 3. CLEAR-SKY FLUX CALCULATIONS Two procedures have been proposed by Cess and Potter [1987] for evaluating clear-sky TOA fluxes in GCMs. In their Method I the clear fluxes are calculated from the subset of grid points in which there is no cloud. Monthly means are then determined directly from this clear subset. While Method I isolates clear grid points in a manner that mimics ERBE clear pixel measurements, the problem is that the subsequentaveraging into monthly means does not incorpo- rate the clear-sky diurnal averaging that is inherent in the ERBE data processing system. The second method, referred to as Method II by Cess and Potter [1987], computes the clear-sky TOA fluxes at each grid point, irrespective of whether it is clear or cloudy, from a second radiation calculation with no clouds present in the atmospheric column (but with all other quantities un- changed). This method thus includes diurnal averaging as in the ERBE data processing. It does not, however, account for the fact that cloudy areas tend to contain more atmo- spheric water vapor than do clear areas, and since enhanced atmospheric water vapor reduces the TOA longwave flux, 20,421